ANALYSIS OF DIET AND FEEDING STRATEGIES WITHIN AN ASSEMBLAGE OF ESTUARINE LARVAL FISH AND AN OBJECTIVE ASSESSMENT OF DIETARY NICHE OVERLAP

Fish larvae and zooplankton were sampled during seven consecutive mont hs from four regions of Wilson Inlet, an estuary in southwestern Austr alia. Mouth size, prey size, and dietary composition oflarvae of the g obiids Afurcagobius suppositus, Pseudogobius olorum, and Favonigobius lateralis, the blenniid Parablennius tasmanianus, and the syngnathid U ro-campus carinirostris were determined. Dietary niche overlap (DNO) w as calculated for co-occurring species pairs, both with and without in corporating a measure of relative prey (zooplankton abundance. Signifi cance of DNO was assessed I) objectively, with bootstrapping of the di etary data and 2) subjectively, by assigning significance to values >0 .6. The diet of A. suppositus was dominated by harpacticoids, polychae te larvae, and the calanoid Gladioferens imparipes, whereas diets of t he other species were dominated by copepod nauplii and postnaupliar st ages of the cyclopoid Oithona simplex, the proportions of the latter i ncreasing with growth of the larvae. Small numbers of large and small prey items were found in the stomachs of A. suppositus (mean=2.5), whi ch had the largest mouth, whereas large numbers (mean= 28.7) of small prey and no large items were found in the stomachs of P. tasmanianus, which had the second largest mouth. Between these extremes, P. olorum, U. carinirostris, and F. lateralis each ate mostly small and intermed iate-size prey, supplemented by a few large prey. The data did not sup port the hypothesis that an increase in the difference in gape size be tween species would decrease the prevalence of significant DNO. The la ck, of a consistent relation between mouth size and DNO among the live species is evidence that interspecific dietary differences reflect di fferences in feeding behavior With bootstrapping, the prevalence of si gnificant (P<0.05) DNO between species pairs was 32.6% when prey data were included in the analyses and 46.5% when prey data were not includ ed. By subjectively assigning significance to DNO values >0.6, we obta ined substantially less conservative estimates that indicated the prev alence of significant I)NO was >53%.